Red Hat Bugzilla – Bug 73242
3 failed install attempts
Last modified: 2007-04-18 12:46:16 EDT
From Bugzilla Helper: User-Agent: Mozilla/4.75 [en] (X11; U; SunOS 5.8 i86pc) Description of problem: 1st installation attempt, anaconda crashed installing from CD#2 2nd attempt, media error on CD#2; I rebooted from CD#1 and ran mediacheck on all 3 CDs -- no errors 3rd attempt anaconda again crashed while installing from CD#2 4th attempt successfully installed SuSE 7.3 to the very same system Version-Release number of selected component (if applicable): How reproducible: Couldn't Reproduce Expected Results: If there was a media problem, I expected mediacheck to find it, but the media passed and anaconda still crashed again. Additional info: Pretty much plain vanilla PC: AMD K6-2/450, 128M RAM, 7.8G Samsung; SiS620 video onboard, Davicom ethernet onboard, CMI sound onboard -- all worked fine during install, and are working fine now under SuSE 7.3. But I can't complete the install of RedHat. Disk partitioning was fine, tweaked with DiskDruid to 47M /boot, 6800M /, 256M swap, 700M /home. Software selection pretty much Workstation install, but not emacs. I have 2 anaconda crash dumps -- hope the bugzilla form lets me attache them.
Created attachment 74425 [details] anaconda crashdump from second failed install
There error messages (there are lots in the log you attached): <4>hdb: irq timeout: status=0xd0 { Busy } <4>hdb: ATAPI reset complete indicate there were errors reading the CD. The mediacheck does not currently accurately duplicate the accesses that occur during an install, and so sometimes the mediacheck will pass but an install will fail.
The root cause was that the IDE CDROM was primary slave -- too much data for the controller to handle. When msf@redhat.com showed me the irq error, I moved the CDROM to secondary slave and the install completed successfully. Thanks for the quick response. As an enhancement, could anaconda be made more resilient in the face of transient errors?